Performance Management in ATM Networks

Abstract

ATM is representative of the connection-oriented resource provisioning classof protocols. The ATM network is expected to provide end-to-end QoS guaranteesto connections in the form of bounds on delays, errors and/or losses. Performancemanagement involves measurement of QoS parameters, and application of controlmeasures (if required) to improve the QoS provided to connections, or to improvethe resource utilization at switches. QoS provisioning is very important for realtimeconnections in which losses are irrecoverable and delays cause interruptionsin service. QoS of connections on a node is a direct function of the queueing andscheduling on the switch. Most scheduling architectures provide static allocationof resources (scheduling priority, maximum buffer) at connection setup time. Endto-end bounds are obtainable for some schedulers, however these are precluded forheterogeneously composed networks. The resource allocation does not adapt to theQoS provided on connections in real time. In addition, mechanisms to measurethe QoS of a connection in real-time are scarce.In this thesis, a novel framework for performance management is proposed. Itprovides QoS guarantees to real time connections. It comprises of in-service QoSmonitoring mechanisms, a hierarchical scheduling algorithm based on dynamicpriorities that are adaptive to measurements, and methods to tune the schedulers atindividual nodes based on the end-to-end measurements. Also, a novel scheduler isintroduced for scheduling maximum delay sensitive traffic. The worst case analysisfor the leaky bucket constrained traffic arrivals is presented for this scheduler. Thisscheduler is also implemented on a switch and its practical aspects are analyzed.In order to understand the implementability of complex scheduling mechanisms,a comprehensive survey of the state-of-the-art technology used in the industry isperformed. The thesis also introduces a method of measuring the one-way delayand jitter in a connection using in-service monitoring by special cells.